Research & notes

How we think about data and evaluation.

Working notes from building training and evaluation data for frontier models, methods we trust, pitfalls we’ve hit, and what turned out to hold up. Grounded in the literature, written from the work.

MethodMay 2026·8 min read

Reasoning traces beat answer-only data

Why capturing the working, not just the result, is what lifts models on the hardest tasks.

EvaluationApr 2026·9 min read

Designing contamination-resistant benchmarks

Building tests that measure capability rather than memorization, and survive being trained on.

RLApr 2026·7 min read

What makes an RL environment trainable

The properties that separate a useful, well-shaped environment from a brittle one.

ProcessMar 2026·6 min read

The 11-criterion review bar, explained

How every item earns its place in a dataset, and what we reject.